PSI - Issue 68

Hugo Mesquita Vasconcelos et al. / Procedia Structural Integrity 68 (2025) 795–801 Hugo Mesquita Vasconcelos et al. / Structural Integrity Procedia 00 (2025) 000–000

799

5

Fig. 5. Left image - background removed images aligned, right image transformed to be aligned with the morning center image

As seen in Fig. 4, there is a visible misalignment between the two thermal acquisitions, which is one of the drawbacks when using UAVs. To address this issue, a transformation matrix was obtained to be applied to the noon image. Initially, a removal background software was used to isolate the blade from the background (Kaleido 2024). Afterwards, Matlab tools were utilized to perform the necessary rotational and translational adjustments to align the images accurately (Matlab 2020). The results of this alignment are shown in Fig. 5. Using the open-source software DJI-Thermal-Decoder (AMDGS 2023), the temperature data for each pixel in the DJI’s thermographic images were extracted. By applying the alignment matrix, the temperature data from both captures were aligned. With the two aligned heat maps, it was possible to perform the differences between the two acquisitions. This difference heat map is presented in Fig. 6. It is visible the average difference of 10 degrees Celsius, a lack of a part of the image on the noon image in the bottom part and a visual difference between the two sides. However, it was also performed a squared difference between the heat maps in order to amplify the differences and make them clearer for the analysis. The quadratic difference image enhances the localized hotter area on the left side of the blade (red rectangle on Fig. 7), indicating temperature variations that suggest the presence of voids or defects, which heat up at different rates. Such voids are commonly found in wind turbine blades due to manufacturing inconsistencies or operational wear. In this case, their position on the leading edge—subject to higher structural stress from water or debris impacts—suggests that this region has sustained significant stress over the blade's extended service life which is congruent to other wind turbines in literature (Guan et al. 2024).

Fig. 6. left - Heat map difference from noon and morning, right – square difference heat map distribution